Slides

Centrality, Rapidity and pT dependence of
isolated prompt photon production in
Pb+Pb collisions with the ATLAS detector
at the LHC
PETER STEINBERG, BROOKHAVEN NATIONAL LABORATORY
QUARK MATTER 2014
MAY 19, 2014
Motivation for inclusive photons
• Photons are a penetrating probe of the hot dense medium
• Do not interact strongly, so are sensitive to the initial state of the
nucleus-nucleus collision
• Important test of our understanding of initial state
• Should scale with nuclear thickness (initial parton flux)
• Sensitive to nPDF effects
2
2
R Pb
g (x,Q =100 GeV )
1.4
FGS10
HKN
1.2
nDS
1
0.8
0.6
0.4
0.2
(EPS09, nDS, HKN, FGS)
EPS09
Salgado et al, arxiv:1105.3919
0
10-5
10-4
10-3
10-2
x
• Potential sensitivity to jet-medium effects
• In principle jets could convert to photons in-medium (Gale et al)
2
10-1
1
Scope of measurement
• This result covers a wide range of nuclear geometry and wide
phase space.
• 4 centrality regions (40-80%, 20-40%, 10-20%, 0-10%)
• 2 pseudorapidity regions (central |η|<1.37 and forward 1.52<|η|<2.37)
• 11 pT regions (logarithmic from 22.1-280 GeV)
!
• Also measure ratio of yields between forward and central η
interval
• Useful to cancel various effects
!
• Yields scaled by the nuclear thickness function TAA and
compared to JETPHOX 1.3.0 NLO pQCD
3
Simulation data samples
• In order to simulate both contributions to the photon
cross section — direct and fragmentation — two
simulated datasets were run
• MC11 PYTHIA 6.4 direct photon (17-280 GeV)
• MC11 PYTHIA 6.4 dijets (17-560 GeV) with hard photon filter
•
O(3 billion) events needed to get 3 million photons
!
• Simulated pp events are merged with real minimum bias
data events, and reconstructed as if they were real data
• “Data overlay” - perfect description of underlying event.
4
Data selection
• Events from 2011 Pb+Pb dataset,
• Integrated luminosity 0.14 nb
(1/Nevt) dNevt/dΣET [1/TeV]
trigged on a 16 GeV
electromagnetic deposition within
0.2x0.1 area in ∆ηx∆ɸ
-1
!
• Minimum-bias event selections used
to clean events
102
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
Minimum bias
16 GeV EM trigger
40 GeV tight photons
γ
-1
-1
LMB
int =5 µ b , Lint=0.14 nb
1
10-2
10-4 40-80%
20-40%
10-20%
0-10%
10-6
• ZDC coincidence, ∆tMBTS < 5 ns,
reconstructed vertex
10-8
0
!
1
2
3
FCal ΣET [TeV]
• Centrality selection performed using
ATLAS forward calorimeter (3.2<|η|
<4.9)
5
Photon reconstruction & shower shape variables
• Underlying event removed in
Cells in Layer 3
∆ϕ×∆η = 0.0245×0.05
Trigge
•
Trigge
Tow r
∆ϕ = 0er
.0982
47
0m
1
Square cells in
Layer 2
∆ϕ = 0
.0245
∆η = 0
.025
m/8 =
4.69 m
m
∆η = 0
.0031
Strip cells in Layer 1
37.5m
η
π0
Figure 5.4: Sketch of a barrel module where the different layers are clearly visible with the ganging
of electrodes in f . The granularity in h and f of the cells of each of the three layers and of the
trigger towers is also shown.
• First layer variables:
•
2
m
00
m
1.7X0
∆ϕ=0.
0245x
36.8m 4
mx
=147.3 4
mm
ϕ
γ
Fraction of cluster ET in hadronic
section
16X0
4.3X0
Containment and width of showers
• Hadronic leakage variables:
r Towe
∆η = 0 r
.1
m
η=0
• Second layer variables:
•
3
2X0
15
∆η=0.1 regions from HI jet
reconstruction
• Photon clusters seeded using sliding
window in 2nd calorimeter layer.
• Photon identification is based on 9
shower shape variables
Hadronic
5.2.2
Rejects candidates with two showers
6
Barrel geometry
The barrel electromagnetic calorimeter [107] is made of two half-barrels, centred around the zaxis. One half-barrel covers the region with z > 0 (0 < h < 1.475) and the other one the region
with z < 0 ( 1.475 < h < 0). The length of each half-barrel is 3.2 m, their inner and outer
diameters are 2.8 m and 4 m respectively, and each half-barrel weighs 57 tonnes. As mentioned
above, the barrel calorimeter is complemented with a liquid-argon presampler detector, placed in
front of its inner surface, over the full h-range.
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
0-10% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
0-10% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
103
102
102
Entries [/0.006]
103
Entries [/0.00025]
Entries [/0.02]
Selecting photons with shower shape variables
103
10
10
1
1
1
0.006 0.008 0.01 0.012
w η,2
103
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
40-80% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
40-80% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
102
10
10
1
1
0.4
0.6
0.8
w s,3
Width
in three strips
-0.1
Entries [/0.006]
102
0.8
w s,3
Entries [/0.00025]
Entries [/0.02]
103
0.6
small shower-shape
corrections applied
to account for
observed data vs. MC
differences
102
10
0.4
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
0-10% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
103
102
0
0.1
Rhad
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
40-80% Central
p =35-44 GeV
T
|η|<1.37
Data
Simulation
10
1
0.006 0.008 0.01 0.012
w η,2
η width in
2nd layer
-0.1
0
0.1
Rhad
Hadronic
leakage fraction
“Tight” selection - satisfies 9 shower shape selections
per photon, tuned on photons in HI events
“Non-tight” selection - Background enhanced by selecting clusters that fail
selections which usually reject having two distinct showers, or two merged showers
7
P(ET,iso)
P(ET,iso)
Isolation selection
0.04
0.04
0.02
0.02
P(ET,iso)
-20
0.08
ET,iso(R=0.3)
-20
40
[GeV]
0.08
0.06
0.04
0.04
0.02
0.02
P(ET,iso)
!
• Simulated isolation distribution,
normalized to data for ET,iso<0, showing
increasing fluctuations in central events
!
• Non-tight photons from data,
normalized for ET,iso>8 GeV, showing
how jets fill the tails in the isolation
distributions
20
0.06
-20
0.1
0
20
-20
40
ET,iso(R=0.3) [GeV]
0.1
0.05
-20
0.15
0
20
40
ET,iso(R=0.3) [GeV]
10-20% Central
0
20
40
ET,iso(R=0.3) [GeV]
0.05
0
20
-20
40
0.15
ET,iso(R=0.3) [GeV]
0.1
0.1
0.05
0.05
-20
0
20
-20
40
ET,iso(R=0.3) [GeV]
8
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
Lint =0.14 nb-1
0-10% Central
p =35-44 GeV
T
|η|<1.37
20-40% Central
P(ET,iso)
• “Isolated”: ET,iso(R=0.3) < 6 GeV
0
P(ET,iso)
depth of the calorimeter in a ring of
R=0.3 around the photon direction,
with photon removed
0.06
P(ET,iso)
• ET,iso is the transverse energy in the full
Tight photons0.08
Non-tight photons
γ Simulation
0.06
P(ET,iso)
0.08
0
20
40
ET,iso(R=0.3) [GeV]
40-80% Central
0
20
40
ET,iso(R=0.3) [GeV]
P(ET,iso)
P(ET,iso)
Double sideband method: ideal
0.08
D
A
0
B
10
0.06
0.04
0.04
0.02
20
30
ET(R =0.3) [GeV]
iso
-20
0.08
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
Lint =0.14 nb-1
0-10% Central
p =35-44 GeV
T
Nontight
|η|<1.37
Tight
B
A
0.02
P(ET,iso)
Tight
C
P(ET,iso)
Non-tight
Tight photons0.08
Non-tight photons
γ Simulation
0.06
20
-20
40
00 E 10(R=0.3)
20
[GeV]
0.08
T,iso
D
C
0
20
40
0
10
20
E (R=0.3) [GeV]
T,iso
Basic principle is straightforward: use non-tight
0.04
photons to extrapolate 0.02
non-isolated/non-tight
photons
0.02
into the signal (tight, isolated region):
10-20% Central
0.06
P(ET,iso)
P(ET,iso)
0.06
of the double side-band approach, showing the two axes for partitioning
photon
the “signal region” (tight and isolated photons) for which efficiencies are defined,
0.04
t, non-isolated photons, region C contains non-tight isolated photons,
and region
nd non-isolated photons.
0
20
-20
40
ible compared with other uncertainties. Hence, no reweighting is-20
applied to the
E (R=0.3) [GeV]
0.1
ccount for any systematic di↵erence between the distributions in 0.1
data and MC, the T,iso
varied in both data and MC as described in Sec. 8.
nt, only the highest pT photon is considered, since far less than 1% of events with
are expected to have a second hard photon, and the overall rate of additional tight,
0.05
ates in the data was found to be less than 1% relative to that for 0.05
the highest-energy
ter application of the tight selection and an isolation criterion of ET,iso < 6 GeV
y sample, there are approximately 51,000 candidates with pT > 22 GeV within
9
andidates within 1.52 < |⌘| < 2.37.
0
20
Yield = A - B (C/D) = A - C (B/D)
(If A/B = C/D, then yield =0)
40
ET,iso(R=0.3) [GeV]
20-40% Central
Non-tight
NN
-1
int
T
P(ET,iso)
P(ET,iso)
iso
P(ET,iso)
T
P(ET,iso)
Tight
nt, only the highest pT photon is considered, since far less than 1% of events with
sigrate of additional tight,
are expected to have a second hard photon, and the overall
A 0.05
661
ates in the data was found to be less than 1% relative to that for the highest-energy
ter application of the tight selection and an isolation criterion of ET,iso < 6 GeV
662
i GeV within
y sample, there are approximately 51,000 candidates with pT > 22
10
andidates withinsig
1.52 < |⌘| < 2.37.
on
P(ET,iso)
P(ET,iso)
n.
652
C and D), then the double sideband approach utilizes
ts
653
the ratio of counts in C to D to extrapolate the measured
ue Double sideband method: actual
654
number of counts in region B to correct the measured
e655
number of counts in region 0.08
A, i.e.
Tight photons0.08
1]
ATLAS Preliminary
Non-tight photons
Pb+Pb s =2.76 TeV
γ
Simulation
ed
L =0.14 nb
obs
C
D
0.06
0.06
N
0-10% Central
sig
obs
obs C
Tight
is
N = NA
NB
(2)
p =35-44 GeV
obs
Nontight
|η|<1.37
0.04
0.04
N
D
ne
B
0.02
0.02
A
A
B
e, 656 Leakage
of signal into the background regions needs to C
D
t” 657 removed before attempting to extrapolate into the signal
-20
0
20
-20
40
0
20
40
0
10
20
30
0
10
20
0
10
20
0.08
ET,iso(R=0.3) [GeV]
0.08
ET,iso(R=0.3) [GeV]
E (R =0.3) [GeV]
n- 658 region. A set of
“leakage factors”, ci , are calculated to
he
10-20% Central
0.06
0.06
of the659
double extrapolate
side-band approach,the
showing
the two axes
forsignal
partitioning
photon in region
number
of
events
A
into
In
reality,
signal
photons
can
“leak” into the sidebands:
the
“signal
region”
(tight
and
isolated
photons)
for
which
efficiencies
are
defined,
ch 660 the other regions.
0.04
0.04
t, non-isolated
photons,
region
C
contains
non-tight
isolated
photons,
and
region
use
simulations
to
estimate
leakage
factors
using
signal-only:
c
X = NX/NA
ch
nd non-isolated photons.
0.02
0.02
⇣
⌘
sig
obs
ns
⇣
⌘ NC
cC NA
obs=Data
sig
sig
obs Hence, no
obsreweighting is-20applied to the
0
20
-20
40 (3) 0
20
40
ible compared withN
other uncertainties.
⇣
⌘
=
N
N
c
N
B
A
B
E (R=0.3) sig
[GeV]
ET,iso(R=0.3) [GeV]
A
A
0.1
ccount for any systematic di↵erence between the distributions in 0.1
data and MC, obs
the T,iso
sig=signal
N
c
N
D
eD
A
varied in both data and MC as described in Sec. 8.
20-40% Central
Solvedfactors
for N are
with calculated
quadratic formula,
and
The leakage
using
the
0.05
sig is zero
sig when AD=BC)
only
one
root
is
physical
(the
one
that
PYTHIA+data sample as c = Ni /NA , where
0.5
0.5
40-80%, |η|<1.37
102
2×102
2
10
2
2×10
30 40
2
10
2
2×10
1
photon p T [GeV]
1
photon p T [GeV]
1
photon p T [GeV]
0.5
0.5
0.5
0.5
40-80%, 1.52<|η|<2.37
0
20-40%, 1.52<|η|<2.37
0
30 40
102
2×102
photon p T [GeV]
102
1
2×102
photon p T [GeV]
30 40
102
A
Nobs
2×102
photon p T [GeV]
0-10%, 1.52<|η|<2.37
10-20%, 1.52<|η|<2.37
0
0
30 4050
A
Nsig
0
0
30 40
P =
0-10%, |η|<1.37
10-20%, |η|<1.37
Purity
30 40
1
ATLAS Preliminary
Pb+Pb sNN = 2.76 TeV
Lint = 0.14 nb-1
0.5
0.5
0
Purity
Purity
1
20-40%, |η|<1.37
0
Purity
1
Purity
1
Purity
Purity
Purity
Double sideband method: purity
30 4050
102
2×102
photon p T [GeV]
30 4050
102
2×102
photon p T [GeV]
Statistical
error assuming
A,B,C,D
follow multinomial
statistics
• The purity extracted from the double sideband method is defined as fraction of
the tight isolated candidates left after removing backgrounds
!
• Beyond dotted lines, statistics in sideband D are small; data extrapolated to
sideband regions, and statistical errors reflect fit errors
11
1
0.8
1
Efficiency
1
Efficiency
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.4
30 40
102
20-40%, |η|<1.37
01
2×102 30 40
2
10
0.2
10-20%, |η|<1.37
01
2
2×10
30 40
2
10
0.2
Efficiency
01
0.2
Efficiency
40-80%, |η|<1.37
photon p T [GeV]
0.8
photon p T [GeV]
0.8
photon p T [GeV]
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0
30 40
102
0
2×102
photon p T [GeV]
20-40%, 1.52<|η|<2.37
30 4050
102
0.2
0
2×102
photon p T [GeV]
0.8
10-20%, 1.52<|η|<2.37
30 4050
102
0-10%, |η|<1.37
01
2
2×10
0.8
0.2 40-80%, 1.52<|η|<2.37 0.2
ATLAS Pb+Pb
Simulation Preliminary
0.8
0.8
0.6
0.2
Efficiency
Efficiency
1
Efficiency
Efficiency
Efficiency in centrality & eta bins
0.2
0
2×102
photon p T [GeV]
30 40
102
2×102
photon p T [GeV]
0-10%, 1.52<|η|<2.37
30 4050
102
2×102
photon p T [GeV]
Relative to PYTHIA photons with ETiso<6 GeV at parton level.
Includes contributions from reconstruction (η dependence), identification (pT
dependence), and isolation (centrality dependence)
No systematic uncertainties shown, since effect of cut variations affect both efficiency and purity. Net effect represented in systematics on yield.
12
Systematic uncertainties
• Variations applied to various parts of the analysis simultaneously
in data and MC, to probe data/MC differences
• Photon selection cuts (shower shape, isolation)
• Effect of leakage correction
• Shower shape corrections removed
• Energy scale & resolution uncertainties
• Removed fragmentation photons from MC
• Evaluated in two regions 22-44.1 GeV, 44.1-111.1 GeV
• Largest of each variation selected and applied symmetrically
• Total uncertainties vary from 10% to 38%, with larger values in
most peripheral events (lower data statistics), forward region
(lower statistics), and central events (larger UE fluctuations).
13
n yields
per-event
yield
val inCorrected
pT , ⌘ and centrality
(C), the per-event
yield of leading photons is define
sig
NA U(pT , ⌘, C)
1 dN
(pT , ⌘, C) =
Nevt (C) d pT
Nevt (C)✏tot (pT , ⌘, C) pT
1. Signal events from sideband
analysis
12
2. Energy scale/resolution corrections U (10% from
22-28.1 GeV, then less than 4% for full
pT and η range)
3. Nevt from full counting of minimum bias events
4. Efficiency from simulations
5. ∆pT bin width (no scaling by ∆η or ∆ɸ)
Contamination from electrons from W estimated to be
only appreciable in 2 bins around 40 GeV:
3.5% correction in central η, 5% at forward η
14
7
Photon yields
Corrected
per-event
yield
vs.
JETPHOX
For each interval in p , ⌘ and centrality (C), the per-event yield of leading photons is defined as
T
sig
10
Central η
106
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
|η|<1.37
JETPHOX 1.3
ET,iso(R=0.3) < 6 GeV
12
(dNγ /dp T)/T AA [pb/GeV]
7
107
Forward η
106
3
Data 0-10% × 10
Data 10-20% × 102
Data 20-40% × 101
0
Data 40-80% × 10
T
(1/N )(dNγ /dp )/T AA [pb/GeV]
NA U(pT , ⌘, C)
1 dN
(pT , ⌘, C) =
Nevt (C) d pT
Nevt (C)✏tot (pT , ⌘, C) pT
ATLAS Preliminary
Pb+Pb sNN=2.76 TeV
1.52<|η|<2.37
JETPHOX 1.3
ET,iso(R=0.3) < 6 GeV
3
Data 0-10% × 10
Data 10-20% × 102
Data 20-40% × 101
0
Data 40-80% × 10
104
evt
104
(4)
102
102
1
1
10-2
10-2
30 40 50
102
2×102
30
photon p T [GeV]
40 50 60 70
102
photon p T [GeV]
15
102
1
0.5
20-40%, |η|<1.37
2
0
Ratio to JETPHOX (pp )
0.5
Ratio to JETPHOX (pp )
Ratio to JETPHOX (pp )
1.5
1
40-80%, |η|<1.37
2
2
10
2
1.5
2
0
2
10
0.5
0-10%, |η|<1.37
2
0
photon p [GeV]
photon p [GeV]
photon p [GeV]
1.5
1.5
1.5
T
1.5
1
0.5
T
1
0.5
40-80%, 1.52<|η|<2.37
1
0.5
20-40%, 1.52<|η|<2.37
0
2
10
0
2
T
T
photon p [GeV]
T
T
0-10%, 1.52<|η|<2.37
10
photon p [GeV]
photon p [GeV]
0.5
0
2
102
1
10-20%, 1.52<|η|<2.37
10
photon p [GeV]
T
ATLAS Preliminary
Pb+Pb s NN=2.76 TeV
Lint = 0.14 nb-1
1
10-20%, |η|<1.37
Ratio to JETPHOX (pp )
0.5
0
JETPHOX PDF+scale err.
JETPHOX Pb+Pb/ pp
JETPHOX EPS09/ pp & err.
1.5
1
2
0
Ratio to JETPHOX (pp )
1.5
2
Ratio to JETPHOX (pp )
2
Ratio to JETPHOX (pp )
Ratio to JETPHOX (pp )
Ratios to JETPHOX
102
photon p [GeV]
T
Ratios to JETPHOX NLO pQCD calculations (CTEQ6.6, BFG II) run with R=0.3/6 GeV isolation. Three configurations: pp (unity), Pb+Pb/pp (black line), EPS09/pp (blue area).
Yellow shaded region is scale & PDF uncertainties, shared with Pb+Pb.
EPS09 errors represented by blue area.
16
0.8
JETPHOX pp
JETPHOX Pb+Pb
0.8
JETPHOX EPS09
0.6
20-40%
Lint = 0.14 nb
0.4
0.2
2
2
10
T
0.6
0.2
0
2
10
photon p [GeV]
T
0-10%
0.4
0.2
0
photon p [GeV]
0.6
0.4
0.2
0
0.8
10-20%
Pb+Pb s NN=2.76 TeV
-1
1
Forward:
1.52<|η|<2.37
0.8
Central:
|η|<1.37
ATLAS Preliminary
0.6
0.4
1
Forward/Central
40-80%
1
Forward/Central
1
Forward/Central
Forward/Central
Forward/Central ratios
0
102
10
photon p [GeV]
T
photon p [GeV]
Ratios of forward and central yields,
compared with pp (yellow), Isospin (black line),
and nPDF EPS09 (blue area).
Total uncertainties are large enough to preclude
strong statements regarding isospin & nPDF effects.
17
T
Conclusions
• Centrality, pseudorapidity and pT dependence of photon
production in Pb+Pb collisions
• First HI measurements at high pT forward η region
• Fully corrected yields
• Background subtracted, efficiency and resolution corrections
• Yields in good agreement with JETPHOX pp
• Comparisons made to Pb+Pb with and without nPDF effects
• Errors still too large to distinguish among scenarios
• Forward/central ratios presented for first time
• Sensitive to isospin and nuclear effects
• Data systematically lower than pp, but the predictions are close
enough that no strong conclusions can be made
18
Extra slides
19
ATLAS
dσ / dET [pb GeV-1]
JETPHOX CTEQ 6.6
iso
ET (∆ R<0.4) < 4 GeV
102
Data 2010
10-1
(a)
|η|<0.6
JETPHOX CTEQ 6.6
iso
ET (∆ R<0.4) < 4 GeV
ATLAS
(b)
data/theory
10-2
1.4
1.2
1
0.8
0.6
100
150
200
250
300
102
Data 2010
350 400
ET [GeV]
∫ Ldt = 35 pb
-1
luminosity uncertainty
10
JETPHOX CTEQ 6.6
iso
ET (∆ R<0.4) < 4 GeV
ATLAS
1
1.4
1.2
1
0.8
0.6
50
dσ / dET [pb GeV-1]
50
100
150
250
300
Data 2010
10-1
350 400
ET [GeV]
∫ Ldt = 35 pb
-1
luminosity uncertainty
10
JETPHOX CTEQ 6.6
iso
ET (∆ R<0.4) < 4 GeV
ATLAS
1
(c)
200
102
1.52≤|η|<1.81
(d)
1.81≤|η|<2.37
10-2
10-2
1.4
1.2
1
0.8
0.6
50
-1
0.6≤|η|<1.37
10-2
10-1
∫ Ldt = 35 pb
luminosity uncertainty
10
1
data/theory
JETPHOX provides
a reliable reference
over the full phase
space in this
measurement
∫ Ldt = 35 pb
-1
luminosity uncertainty
10
10-1
data/theory
35
of 7 TEV
data show quite
good agreement with
JETPHOX (with
a tighter isolation)
Data 2010
1
dσ / dET [pb GeV-1]
pb-1
102
data/theory
JETPHOX vs. ATLAS data
dσ / dET [pb GeV-1]
Phys.Lett.B 706 (2011) 150-167
100
150
200
250
300
350 400
ET [GeV]
1.4
1.2
1
0.8
0.6
50
100
150
200
250
300
350 400
ET [GeV]
Figure 2: Measured (dots) and expected (shaded area) inclusive prompt photon production cross-sections, and their ratio, as a function of the photon ET and in the
range (a) |⌘| < 0.6, (b) 0.6  |⌘| < 1.37, (c) 1.52  |⌘| < 1.81 and (d) 1.81  |⌘| < 2.37. The data error bars combine the statistical and systematic uncertainties, with
the luminosity uncertainty shown separately (dotted bands).
20